基于IRT不同参数估计方法的考生能力估计结果的比较  

Comparison of Examinee’s Ability Estimation Results Based on the Different IRT Estimation Methods

作　　者：马洪超[1] 

机构地区：[1]北京语言大学汉语速成学院

出　　处：《考试研究》2012年第1期61-66,85,共7页Examinations Research

摘　　要：参数估计是项目反应理论应用、发展的前提。本研究针对六种不同的HSK考生样本,分别使用三种软件,采用不同的参数估计方法对考生能力值进行估计,结果表明能力值估计结果与考生潜在能力分布有关系。当潜在能力分布趋向正态分布时,能力值的估计的误差较小。此外,不同软件的参数估计方法的能力值估计结果均有差异。Parameter estimation is the prerequisite for the application and development of IRT. Based on six samples of HSK examinees, examinees＇ ability are estimated by using three different software programs with different estimation methods. It shows that estimation result of ability is related to the distribution of latent trait. When the latent trait is closed to normal distribution, the error of ability estimation is smaller. The results of ability estimation are different when the programs and estimation methods are different.

关 键 词：参数估计 最大似然估计(MLE) 后验期望估计(EAP) 后验众数估计(MAP) 

分 类 号：G424.74[文化科学—课程与教学论]